Re: [PATCH 2/3] ARM: dts: rk3288 Tinker Board (S) add wi-fi

[Tony McKahan <tonymckahan-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]

Hello David,

  I agree, of course coordination is not always so simple, I did not
know Arch didn't have this configured yet.  As for patching, I agree,
but the Tinker has some... peculiarities, I am a hardware guy,
software is a passtime, so my code is not always perhaps up to
standard.  And I wasn't 100% on submitting changes to the device tree
based on a staging driver.  Most of my work has been absorbed into
https://github.com/Miouyouyou/RockMyy .  As for your approach, as I
said, the wifi is the same as I have except for the reset/enables and
the DDR50, which I didn't know the rtl8723bs supported.  BT I have no
in-kernel support for other than enabling the RTS pin to the device
tree, I handle it in user space for now.  I've been watching the
hci_h5 conversation, I'll look into testing these changes/options
tonight perhaps once I make sure our rockchip-dev recipes are still
valid with the newest RC, and can have them committed so others can
test.

Thank you,
-Tony

On Sun, Feb 17, 2019 at 10:55 AM David Summers
<beagleboard-8lkWs+DogrrqRUWtDFdff/XRex20P6io@public.gmane.org> wrote:
>
> On 17/02/2019 14:32, Tony McKahan wrote:
> > A similar configuration has been in use on Armbian images since kernel
> > 4.14, but not in a clean enough format for submission. I will test
> > this, but the only differences I see compared to ours are combination
> > of the mystery "chip_h" power domain/reset with the Wifi one and the
> > inclusion of sd_uhs_ddr50.
> >
> > -Tony
> >
> > On Sun, Feb 17, 2019 at 7:16 AM David Summers
> > <beagleboard-8lkWs+DogrrqRUWtDFdff/XRex20P6io@public.gmane.org> wrote:
> >> This patch adds the wifi to the ASUS Tinker Board (S) machines.
> >>
> Tony, thanks for the interest in this and the Bluetooth RFC. To my mind
> it would be good if we could mainline an agreed solution. As much as
> anything doesn't make sense to me for Armbian and ArchLinux Arm to have
> independent solutions.
>
> In the ArchLinux Arm  camp I'm not a fan of applying many patches to
> support a board, as much as anything it makes that board non generic -
> and so has to have its own hand crafted kernel. This to me isn't a good
> solution, so if possible prefer the pain of getting it mainlined.
>
> Stefan Wahren has come up with a minimal patch to hci_h5 for bluetooth,
> that says the relevant properties can be read from the device tree. This
> means that acpi is no longer necessary. Thats currently out in testing.
>
> Anyway would be good if Arch and Armbian could merge our efforts ...
>
> Regards,
>
> David.
>